Display VNF monitoring status and health failure reason

[Sridhar Ramaswamy]

Context:

Tacker monitoring feature allows a monitoring-driver to check the health of VNF and report or heal (using a respawn) on a health event. Currently only way to figure out health monitoring is active on VNF is by looking at the logs!! And when a health condition occurs the only way the operator get to know the situation is by VNF status flipping from ACTIVE --> DEAD --> back to ACTIVE.

We need better visual representation of the ongoing monitoring and, even failure occurs, the failure reason should be displayed.

Request for Enhancement:

Enhance vnf-show API to report,

a) whether monitoring is enabled for the VNF (even for one VDU for a multi-VDU VNF)
b) whether monitoring driver is successfully called / working
c) On a failure event, vnf-show should display the last-seen-failure-reason (like VNF unreachable, call capacity reached, etc)

(a) and (b) should be used in Tacker Horizon GUI to show monitoring being active for a VNF (may be a beating heartbeat gif) and a new column to display VNF event.

Sample CLI / API,

$ tacker vnf-show myvnf

...
  monitoring : { enabled: true, active: true }
  monitoring-action-reason : { "VNF unreachable, respawn" }
...

[Sridhar Ramaswamy (srics-r)]

I believe (c) will be taken care by the proposed blueprint,

https://blueprints.launchpad.net/tacker/+spec/audit-support

However (a) and (b) would be still worth displaying.

[Sridhar Ramaswamy (srics-r)]

Bharath - are you planning to work on this ? If not, can you please un-assign so that someone else from the community could pick this up.

[Tung Doan (tungdoan)]

"when a health condition occurs the only way the operator get to know the situation is by VNF status flipping from ACTIVE --> DEAD --> back to ACTIVE" --> well said!

Hi Sridhar,
I would like to add one more item to the list. It would be nice to have "state" field as mentioned in my spec: https://review.openstack.org/#/c/434974/2/specs/pike/ns-lifecycle-mgmt.rst@85

Accordingly, operators should know if vnf is respanwed or not. The transition (ACTIVE --> DEAD --> back to ACTIVE) is not enough.